14-18 Hirsch-Hartmann SGA-OEGA 2013 OK

27
Persistence of firm-level profitability in the

European dairy industry
S. Hirsch and M. Hartmann1
Abstract - Based on autoregressive (AR) models and

Arellano-Bond dynamic panel estimation this article
analyses profit persistence in the European dairy
industry. The sample comprises 590 dairy processors
from the following five countries: Belgium, France,
Italy, Spain and the UK. The AR models indicate that
cooperatives which account for around 20% of all
firms in the dairy sector are not primarily profit oriented. In addition, the results point towards a high
level of competition as profit persistence is rather low
even if cooperatives are excluded. The panel model
reveals that short as well as long run profit persistence is influenced by firm and industry characteristics. 1
INTRODUCTION
While the competitive environment hypothesis postulates that firm profits which deviate from the competitive norm cannot exist in the long run, such
persisting abnormal profits are rather the normal
case in the real world. Although there is a great
quantity of studies analyzing profit persistence in
entire manufacturing sectors (e.g. Mueller, 1986;
Gschwandtner, 2005), empirical evidence for the
European food industry and its subsectors is still
scarce. Therefore, this paper tries to fill this gap by
analyzing the phenomenon of profit persistence in
the European dairy industry. The study is based on a
large sample of 590 European dairy processors.
THE MODEL
Starting with Mueller (1986) the simple autoregressive process of order one AR(1) has become the
econometric cornerstone of the empirical profit persistence literature. The AR(1) is a simple regression
of firm is abnormal profit at time t ( i,t ) on the
immediate previous level:
i ,t i i i,t 1 i ,t
in order to decide which model describes the adjustment path best.

Equation 1 yields two profit persistence measures.
The first one is i which indicates the speed of convergence of profits to the long run level. Since
also reflects the fluctuations in profits, it can be

interpreted as short run persistence. Small (Large)
values of i imply that competitive forces on firm i
are rather strong (weak) while profit persistence is
low (high). In the literature the mean value of i
across all analyzed firms has become the main
measure for persistence. The second measure is
long run persistence. It is reflected by the long run
average
of
the
autoregressive
process
p i i /(1 i ) . p i is a measure of permanent

rents, which are not eroded by competitive forces in
the long run. The percentage of p
i s significantly
different from zero in a given sample can therefore
be interpreted as an additional indicator of the degree of persistence within it.
In order to explain the persistence measures i
and
p i
the majority of previous studies implements
a second estimation step where several firm and

industry characteristics are regressed on i and p
i .
However, as this method is plagued by econometric
flaws (Baltagi, 2008) a dynamic panel model according to equation (2) was estimated instead, using
Arellano and Bonds GMM estimator.
i , t j ( X j ,i ,t ) j ( X j ,i , t ) i ,t 1 i ,t
j
(2)
The ( X s) are specific firm and industry characj
(1)
teristics that are expected to influence profit persistence. The impact of the X s on short-run persisj
is a white noise error term with zero
tence can be evaluated by the s. The j s reflect

j
mean and constant variance. We extent this approach according to Gschwandtner (2005) by estimating up to four lags for each firm and afterwards
using Schwarz-Bayesian Information Criterion (SBC)
the impact of the X s on abnormal profits over the

j
where
i,t
entire time period analyzed and it can be assumed

that the direction of this impact prevails in the longrun. It is therefore possible to assess the direction of
change in long-run profit persistence for a given
change in the variables X by means of the algebraj
Stefan Hirsch and Monika Hartmann are from the Institute for Food
and Resource Economics (ILR), University of Bonn, Germany ([email protected]).
ic signs of the j s.
28
DATA AND RESULTS

Firm level data was taken from AMADEUS, a commercial balance sheet database while industry data
is constructed from the Eurostat database. This
study is based on the 13 year period 1996 through
2008 since this is the longest available for the European dairy industry. Abnormal firm profits ( i,t ) are
measured by return on assets (ROA) in year t normalized by mean ROA of that year as a proxy for the
competitive norm. The screened sample contains
590 firms from Belgium, France, Italy, Spain and the
UK.
Table 1 shows that the mean value of i for the
pact on short- and long-run persistence. This result

is not surprising as high retailer concentration likely
leads to strong bargaining power, putting dairy processors under pressure.
Table 2. Dynamic panel estimation of equation 2.
Coeff. ( )
Variable
Coeff. ( )
MS*i,t-1
0.001**
MS
-0.000**
Age*i,t-1
-0.045***
Age
Variable
Ln TA*i,t-1
0.136*
Ln TA
0.007
Gr. TA*i,t-1
0.006**
Gr. TA
0.001**
-0.000*
Gear
-0.000*
dairy industry is 0.094 which is rather low compared

to other manufacturing sectors and to the entire
European food industry. Hirsch and Gschwandtner
(2013) find for the entire food industry mean values
for i of 0.057 in Belgium, 0.143 in Italy, 0.188 in
1/Curr*i,t-1
-0.408***
1/Curr
-0.023***
HHI*i,t-1
9.060***
HHI
0.177
NF*i,t-1
-0.605
NF
-0.092***
-2.211***
Gr. NF
0.014
R&D*i,t-1
-1.961***
R&D
0.024
France, 0.201 in Spain and 0.232 in the UK.

If cooperatives which are not mainly profit oriented are excluded from the sample, the mean i re-
CR5*i,t-1
-11.925***
CR5
-0.137***
sembles the ones for the entire food industry.

About 41% of the firms in the sample are estimated to earn long- run profits that deviate from the
competitive norm. However, only one quarter of
these firms is earning long-run profits above the
competitive norm while all significant long-run values for cooperatives are negative. Without cooperatives 32% of the firms are achieving abnormal profits in the long run. This value is, however, lower
than in the entire European food industry where
Hirsch and Gschwandtner (2013) find values of
around 40%.
Table 1. An overview of the persistence parameters.
Mean
All firms
Coop.b
All otherc
0.094
-0.184
0.163
s sign.a 0
% of
41.2
77.3
32.2
s sign. >0
% of
9.8
0.0
12.3
s sign.a <0
% of
31.4
77.3
b
19.9
c
significant at the 5% level or less; Cooperatives; All other

legal forms except cooperatives
The results of the dynamic panel estimation in

Table 2 show that firm size and firm growth have a
positive impact on profit persistence while age and
R&D have a negative impact on the short-run value.
The negative impact of age can be attributed to a
corporate aging problem with organizational rigidities, slower growth and assets, which become obsolete with time (Loderer and Waelchli, 2010) while
the negative impact of R&D could be a consequence
of the fact that innovations in the dairy industry are
in most cases only minor product extensions. The
consistently negative impact of firm risk contradicts
standard risk theory. However, a negative risk-profit
relationship, known as Bowmans (1980) risk-return
paradox', is also a long-established fact and could
explain the present results.
While the degree of concentrations within the
industries in which the dairy processors operate has
a positive impact on short-run persistence, concentration in the food retail sector has a negative im-
Gear*i,t-1
Gr. NF*i,t-1
Wald
Hansen
AR(2)
(21) = 222.86*** p=0.000

(36) = 45.48 p=0.134
z = -1.61 p=0.108
Dependent variable: i,t (abnormal profit)

Firm variables: MS = firm sales/industry sales; Age = firm
age; Ln TA = natural logarithm of total assets; Gr.TA=
growth rate of total assets; Gear = gearing ratio; 1/Curr =
1/current ratio. Industry variables: HHI = HerfindahlHirschman Index; NF = number of firms in industry divided
by industry sales; Gr.NF = Growth rate of NF; R&D = Share
of R&D expenditure in industry value added; CR5 = Fivefirm concentration ratio of the retail sector.
a
Age was dropped from the model due to multicollinearity.
***, **, *significant at the 1%, 5%, 10% level respectively.
CONCLUSIONS
To summarize, the results show that profit persistence in the dairy industry is essentially lower compared to other sectors outside the food industry. The
low mean i s in the food industry can be attributed
to a high degree of market saturation, strong price
competition and a highly concentrated retailing sector whose bargaining power is even strengthened by
a high and still increasing share of private labels.
Similar to previous results for the whole food industry firm size has a positive impact on profit persistence while R&D has a negative effect.
REFERENCES
Baltagi, B. H. (2008). Econometric Analysis of Panel
Data, 4th ed. Chichester: John Wiley & Sons.
Bowman, E. H. (1980). A Risk/Return Paradox of
Strategic Management. Sloan Management Review
21(3):17-31.
Gschwandtner, A. (2005). Profit persistence in the
very long run: evidence from survivors and exiters.
Applied Economics 37:793-806.
Hirsch, S. and Gschwandtner, A. (2013). Profit Persistence in the Food Industry: Evidence from five
European Countries. European Review of Agricultural
Economics, forthcoming.
Loderer, C. and Waelchli, U. (2010). Firm age and
performance. Working Paper, University Bern.
Mueller, D. C. (1986). Profits in the Long Run. Cambridge: Cambridge University Press.

14-18 Hirsch-Hartmann SGA-OEGA 2013 OK

Uploaded by

Copyright:

Available Formats

14-18 Hirsch-Hartmann SGA-OEGA 2013 OK

Uploaded by

Document Information

Original Description:

Copyright

Available Formats

Share this document

Share or Embed Document

Sharing Options

Did you find this document useful?

Is this content inappropriate?

Copyright:

Available Formats

14-18 Hirsch-Hartmann SGA-OEGA 2013 OK

Uploaded by

Copyright:

Available Formats

27

Persistence of firm-level profitability in the

Abstract - Based on autoregressive (AR) models and

in order to decide which model describes the adjustment path best.

also reflects the fluctuations in profits, it can be

p i i /(1 i ) . p i is a measure of permanent

the majority of previous studies implements

a second estimation step where several firm and

The ( X s) are specific firm and industry characj

is a white noise error term with zero

tence can be evaluated by the s. The j s reflect

the impact of the X s on abnormal profits over the

entire time period analyzed and it can be assumed

DATA AND RESULTS

pact on short- and long-run persistence. This result

dairy industry is 0.094 which is rather low compared

France, 0.201 in Spain and 0.232 in the UK.

sembles the ones for the entire food industry.

significant at the 5% level or less; Cooperatives; All other

The results of the dynamic panel estimation in

(21) = 222.86*** p=0.000

Dependent variable: i,t (abnormal profit)

You might also like